This paper brings life to old data from which consistent formulations of the molecular and radical channel quantum yields in CHDO photolysis are extracted. The results probably merit publication after considering a number of comments.

In general, the Figure captions appear to be of the “insider type” and not such that the Figures can be viewed and understood by the reader without a meticulous inspection of the text. The criticism applies to most of the figures; taking Figure 1 as example: "Comparison of the measured data by Miller and Lee (1978) (full dots) with the pressure dependent part of the fitted function fkt(M]) for different wavelengths in nm as indicated." Without consulting the text, the reader is left wondering: Which data? Which function? What is Q(M)?

Table 1 is not "self-consistent" in the sense that all symbols are defined either in the Table caption or in footnotes (De1 and De2).

The first Figure referenced in the text is Figure 4a in Section 3 (line 106). It is as if the authors have done a last-minute paragraph swap. Caption to Figure 4a: "Wavelength dependence of the contributions of the 3 terms of equation F7 to the total quantum yield of the CHDO photolysis at 10 hPa (a) and 1030 hPa (b)." At this point in the text, the reader is about to be introduced to where the rate coefficients k₁, k₂, k₅ and k₆ origin (that is, Table X in the paper by Miller and Lee, 1978). The rate coefficients and the derived rate coefficients (k₃, k₄, k₆ and k₇) are then summarized in Table 2. The authors add confusion by citing Table 2 of Miller and Lee (it is actually TABLE II) and then their own Table 2 a few lines later.

Equation (F4) is a pseudo-equation that does not add to the value of the paper and it would perhaps be better write equation (F3) in the form below and maybe indicate the wavelength dependency as well:

Table 2. The excitation wavelengths listed are rounded numbers from the Miller and Lee (ML) 1978 publication that states: the out-of-plane bending mode (n₄) is the inducing mode for radiative transitions and also the promoting mode for the nonradiative transitions. Why not use the wavelengths given by ML ? Why are the 316.2 nm ML data not used? The ML 320.2 nm data are entered in Table 2 with 329 nm excitation – why? Have the authors located misprints in the ML TABLE II?  If yes, this should be communicated. The 329, 344 and 353 nm data for k₅ and k₆ in Table 2 do not correspond to any t-numbers in the ML TABLE II. Where do these numbers come from? The 329 nm data (ML 320.2 nm) refers to fluorescence from the 1¹4¹ (S₁) state (n₁ corresponds to the CH str), and the “next lower vibrational level” would be the 4¹ (S₁) state, for which to fluorescence data are found for 353 nm excitation. However, this is the only case not involving the n₂ mode (C=O str). Obviously, there are no data for the “next lower vibrational levels” of the 4³ and 4¹states involving n₂, yet numbers appear for k₅ and k₆ in Table 2. In summary, the selection of data could be better described.

Line 154. The authors state: The pivot wavelength 1/ε₀ is 348.6 nm, as published in Nilsson et al. 155 (2014). Maybe they should mention that the 348.8 nm origins from quantum chemistry calculations of the barriers to dissociation H-CHO, H-CDO, D-CHO and D-CDO.

Line 168. “optimal values” should probably be “optimized values”

Table 3, Line 185. The A-value for k₁ must have been derived taking the k₁-value for 329 nm as an outlier. Are there other examples of data-massage? There are no error estimates included in Table 3 to indicate the validity of the smoothing procedure.

Line 242. The numbers in the equation are not the same as those given in the reference, but are apparently rounded. However, some of the numbers are rounded outside the error limits given in the reference.

Figure captions, Line 562. As mentioned, the Figures including captions should preferably be self-consistent. In general, this is not the case.

Reviewer`s remarks to egusphere 2023-2254 (Röth and Vereeken)

The photolysis of formaldehyde is a major source of hydrogen throughout the atmosphere. Since the photolysis rates of formaldehyde and its deuterated analogues, pimarily CHDO, in their molecular channels are altitude dependent, there is an isotopic enrichment of HD relative to H2 in the stratosphere. Quantification of this enrichment is a long standing issue in atmospheric science. Based on previous experimental data on the fluorescence as well on the photo-decomposition of formaldehyde and its deuterated analogues, together with detailed molecular kinetic  modelling, the present paper provides an improved understanding of the wavelength dependence of molecular and radical channels of the photolysis of formaldehyde which goes beyond previous interpretations. The paper therefore is valuable and hence publishable povided that a number of aspects are taken into account:

• In the summary stronger emphasis of the present findings with respect to their atmospheric relevance should be made in order to attract stronger interest of the average readership of this journal.
• The organisation of the manuscript has a number of deficits which should be rectified. The first one is the presentation of the overall mechanism to include both radiative as well as collisonal deactivation steps. For instance: There is energy transfer (quenching) allowed in both S₁ and S₀ electronic states. But why is no fractionation of CHDO* – either via a molecular or radical channels - from the electronically excited S₁ state included? Is the initial energy insufficient? The fact that the radical channels are pressure independent, as stated later in the manuscript,  should be indicated in the mechanism. In the form of the mechanism presented here,  reactions (2a), (3a) and (6a) imply that this is not the case. The inclusion of a schematic energy diagram of both eletronic states involved and their threshold energies for decomposion together with arrows for the different pathways of excitation und de-excitation via the cascade would significantly improve the understanding of the mechanism. In such mechanisms with energy dependent channels the fundamental molecular rate coefficients are principally k(E)s as delineated in detail in the paper by Troe. These become thermal rate coefficients k only with the underlying assumption that the system is thermalized at all energies. Very likely this is the case under atmospheric conditions, but it should be mentioned. In the form pesented in the manuscript, all quenching reactions are not mass conserving because the collider M does not  show up on the product side of the equations. Please correct. It should also be mentioned that due to this mechanism and the consecutive reactions of the products only the molecular channel contributes to isotopic fractionation. The products of the radical channel (H+CHO) never end up as H2.
• On line 262 it is stated that the bond strength of C-H and C-D are almost the same. In the view of the reviewer this is not consistent with the notion that the zero point energy for C-D is lower than that of C-H. It also contradicts the statement in this paper that the „threshold energies for the radical channels in CH2O and CHDO are different (lines 271-273)“
• The correspondence between text, figure captions and figures needs much better organisation in order to improve the readability of the manuscript. For instance, Fig. 4 should never be the first figure to be cited in the text. Moreover, the figure captions should be more self-explanatory and more cmprehensible even without recurrence to the rext.
• Improvements of minor formalities:

In line 266 is a „neither“ missing

In line 275 replace „1-term“ by „one-term“

The chapter headline „Isotopic fractionation during the photolysis of CH2O“ should be extended to include CHDO

Line 606-608 figure caption of Fig. 11 needs rephrasing.